Developing automation to suit a range of market environments has put MarketAxess’s clients ahead in the drive for more efficient trade execution and adaptable trading strategies.
MarketAxess has launched Adaptive Auto-X to support a range of trading workflows to help traders benefit from automation in all market conditions. Gareth Coltman, global head of trading automation at MarketAxess, tells us how the product is changing the way trading desks deliver best execution.
What were the market drivers for development of Adaptive Auto-X?
Market structural changes point towards an increasing velocity of trading in fixed income. The indicators include increasing indexation, decreasing capacity of dealer balance sheet, increasing speed of trading, which lead to a larger number of smaller tickets in the marketplace and driving down the average size of trades. All of these qualities set the market up to operate in a higher velocity, more systematic way.
We’ve seen the sell side respond by deploying technology for systematic market making, with systematic firms deploying their own algos to service their exchange traded fund (ETF) business by providing liquidity directly to clients.
Client feedback centres around supporting more efficient execution workflows, with a significant trend towards increasing no-touch activity from clients wanting to maximise that efficiency. There’s also a desire from clients to find ways to interact with a wide range of liquidity, beyond traditional liquidity providers and traditional RFQ-based liquidity. This can be tapping into natural liquidity, be that other clients or dealers offering natural liquidity directly to clients or to tap into order books, regardless of whether those are institutional or retail flows. Clients can see there’s available liquidity out there, and need a workflow to tap into it with an efficiency level that operates effectively.
These two trends drove us to extend the technology that we’d already built to provide RFQ automation for the buy side into something more sophisticated – Adaptive Auto-X.
We’re taking unique predictive AI models that we’ve built, and combining them with the breadth of the liquidity offering that is available on MarketAxess, and ultimately using automation to allow clients to interact with those models in an efficient manner.
Are you also supporting dealers in their efforts to automate more?
We do also offer a range of support to the sell side that lets them create the same kind of systems. We offer a range of AI-driven data products, which allows dealers to make their own modelling and predictive analytics, plus a full suite of API’s that lets them access execution services on the Open Trading Network, which they have increasingly been using to manage risk.
Dealers want a different proposition from the buy side, and they’re more likely to self-build. We’ve seen some buy side firms take a similar approach. We built automation and data products in a box concept, but a small number of buy side firms have chosen to self-build their own EMS-type products in house.
How can you successfully support both models?
We’re not telling the buy side or the sell side how they should interact. We offer a range of solutions all the way from API’s and data feeds to allow self-build through products like Adaptive Auto-X, which are designed to be used systematically, and effectively provide an automated systematic trading solution without having to build it yourself. We also integrate these data products and most of our automation suite directly into our front end. It is a menu-based approach allowing clients and dealers to choose how they’re going to interact with our services.
How have market conditions impacted the pace of progress?
We’ve had some very unusual market conditions over the last few years, particularly this year, post the banking crisis. For example, ETF liquidity provision in a very low volatility environment is generally lower, so you will see less systematic provision, leading to clients to look at other mechanisms for trading. The bulk of the market is still algorithmic, but relative to a year ago when the market was more volatile, it may have fallen.
With that in mind, in which circumstances will Adaptive Auto-X be most useful?
As its name suggests, Adaptive Auto-X can potentially suit any markets. It is intended to be a hands-off algorithmic type approach, where the algorithm is capable of making decisions about the right size, time, protocol and liquidity pool to go to, depending on your objective. For your execution objective, you can parameterise Adaptive Auto-X to be as constrained as you want it to be but ultimately, the algo executes against any liquidity that is available. It can aggress the market when a client wants to trade with urgency, or behave in a passive way, and sweeping up any natural liquidity that’s available to create price improvement for a client.
How do you see client use evolving?
At a high level, it offers a lot of different types of execution workflows, but the main parameters are time horizon and the urgency to execute a trade. Are they benchmarked against arrival or more interested in achieving maximum price improvement? Or minimising market impact? The investment objective and execution benchmarking will affect how a client will interact with the tool. We’re live with a couple of very flexible algos, one of which is focused on passive liquidity seeking, designed for a client who has time to work an order and benefit from the price improvement that natural matching can provide. We also have a smart RFQ algo designed for situations where an order needs to be completed within a certain amount of time. It integrates passive liquidity seeking, along with being able to aggress the market via RFQ or order book, to try and achieve a client’s objective in terms of urgency, while maximising price improvement.
In which ways is this offering unique to MarketAxess?
The liquidity network and the range of trading protocols we have is broader than any other platform, meaning there’s more opportunity for this tool to be very powerful. There’s more price improvement available paired with our award-winning AI driven analytics. So the secret sauce lies in the liquidity and the data products we offer. Our recent acquisition of Pragma really cements our tech stack, which is future-proofed around fixed income automation, because it’s been developed and proven in much higher-frequency, high velocity asset classes.
What is next?
As we’re coming to the end of a very successful pilot with a number of our most significant clients, the feedback has been excellent. We’re now focused on integrating other protocols and liquidity sources into the algo. The goal is to make it even more intelligent by adding additional AI-based analytics, which will make accessible any available source of liquidity in the market today, including streaming liquidity, and ultimately to extend out the offerings to other markets like Europe.
©Markets Media Europe 2023
